Your Next Publishing Product Should Be Aimed at AI Agents
summary
Argues publishers must build for AI agents by running an MCP server with OAuth authentication that mediates access to publisher knowledge bases. Demo uses four Project Gutenberg volumes chunked at ~10,000 characters with ~10% overlap, embeddings from AWS Titan/Bedrock, stored in Mark Logic with a Mark Logic vector search, and served to Claude Desktop via an MCP remote proxy. The MCP enforces login, returns scored JSON chunks, and the agent correctly diagnoses examples (botfly plus two unexpected matches), showing an end‑to‑end, working monetizable flow; speaker cites an estimated $2B annual publisher loss.
AI-written summary · about
chapters (8)
- 0:00Introduction and motivation
- 2:22Business case for MCP servers
- 7:00Technical components overview
- 9:11Live demo of MCP in action
- 12:26Setting up the knowledge base
- 18:45Building the MCP server
- 29:48Configuring the MCP client
- 32:40Q&A and closing
related talks
45mAgents, Tools, and MCP, oh my! Next-level AI concepts for developers
53mAgents, Tools, and Mcp, Oh My! Next Level AI Concepts for Developers
36mPutting AI Agents to Work: A New Era of Open Source Connectivity with MCP
30mComposable AI: Building Next-Gen AI Agents with MCP
28mThe Agentic Developer: Orchestrating AI Workflows with MCPs
40mBuilding AI Agents with Spring & MCP
21mBuilding Agents (the hard parts!)
20mAI Agent Skills and MCP: Better Together